lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   4 ;;   2007, 2008, 2009, 2010, 2011 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  83    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  84    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  85    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  86    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  87    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  88    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  89    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  90    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  91    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  92    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  93    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  94    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  95    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  96    ])
  97
  98 (defun shuowen-radical (number)
  99   (aref shuowen-radicals (1- number)))
 100
 101 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 102
 103 (defvar char-db-ignored-attributes '(ideographic-products))
 104
 105 (defvar char-db-coded-charset-priority-list
 106   '(ascii
 107     control-1
 108     latin-iso8859-1
 109     latin-iso8859-2
 110     latin-iso8859-3
 111     latin-iso8859-4
 112     latin-iso8859-9
 113     latin-jisx0201
 114     cyrillic-iso8859-5
 115     greek-iso8859-7
 116     thai-tis620
 117     =jis-x0208
 118     =jis-x0208@1978
 119     =jis-x0208@1983
 120     japanese-jisx0212
 121     chinese-gb2312
 122     =jis-x0208@1990
 123     chinese-cns11643-1
 124     chinese-cns11643-2
 125     chinese-cns11643-3
 126     chinese-cns11643-4
 127     chinese-cns11643-5
 128     chinese-cns11643-6
 129     chinese-cns11643-7
 130     =jis-x0213-1
 131     =jis-x0213-1@2000
 132     =jis-x0213-1@2004
 133     =jis-x0213-2
 134     korean-ksc5601
 135     chinese-isoir165
 136     katakana-jisx0201
 137     hebrew-iso8859-8
 138     chinese-gb12345
 139     latin-viscii
 140     ethiopic-ucs
 141     =big5-cdp
 142     =gt
 143     =adobe-japan1-0
 144     =adobe-japan1-1
 145     =adobe-japan1-2
 146     =adobe-japan1-3
 147     =adobe-japan1-4
 148     =adobe-japan1-5
 149     =adobe-japan1-6
 150     =hanyo-denshi/ja
 151     =hanyo-denshi/jb
 152     =hanyo-denshi/jc
 153     =hanyo-denshi/jd
 154     =hanyo-denshi/ft
 155     =hanyo-denshi/ia
 156     =hanyo-denshi/ib
 157     =hanyo-denshi/hg
 158     =hanyo-denshi/jt
 159     =hanyo-denshi/ks
 160     =daikanwa
 161     =daikanwa@rev2
 162     =daikanwa@rev1
 163     =cbeta
 164     =gt-k
 165     ideograph-hanziku-1
 166     ideograph-hanziku-2
 167     ideograph-hanziku-3
 168     ideograph-hanziku-4
 169     ideograph-hanziku-5
 170     ideograph-hanziku-6
 171     ideograph-hanziku-7
 172     ideograph-hanziku-8
 173     ideograph-hanziku-9
 174     ideograph-hanziku-10
 175     ideograph-hanziku-11
 176     ideograph-hanziku-12
 177     =>>>jis-x0208
 178     =>>>jis-x0213-1
 179     =>>>jis-x0213-2
 180     =>>jis-x0208
 181     =>>jis-x0213-1
 182     =>>jis-x0213-1@2000
 183     =>>jis-x0213-1@2004
 184     =>>jis-x0213-2
 185     =>>jis-x0208@1978
 186     =>>hanyo-denshi/ft
 187     =>>hanyo-denshi/ks
 188     =>>gt
 189     =>>daikanwa
 190     =+>jis-x0208
 191     =+>jis-x0213-1
 192     =+>jis-x0213-2
 193     =+>jis-x0208@1978
 194     =>jis-x0208
 195     =>jis-x0208@1997
 196     =>jis-x0213-1
 197     =>jis-x0213-1@2000
 198     =>jis-x0213-1@2004
 199     =>jis-x0213-2
 200     ==>ucs@bucs
 201     =>ucs@iso
 202     =>ucs@unicode
 203     =>ucs@jis
 204     =>ucs@JP
 205     =>ucs@jis/1990
 206     =>ucs@cns
 207     =>ucs@ks
 208     =+>ucs@cns
 209     =+>ucs@ks
 210     =>>ucs@iso
 211     =>>ucs@unicode
 212     =>>ucs@jis
 213     =>>ucs@cns
 214     =>>>ucs@iso
 215     =>>>ucs@unicode
 216     =ucs@iso
 217     =ucs@unicode
 218     =>>big5-cdp
 219     =>>gt-k
 220     =>gt
 221     =>big5-cdp
 222     =>daikanwa
 223     =big5
 224     =big5-eten
 225     =>gt-k
 226     =zinbun-oracle
 227     =>zinbun-oracle
 228     =ruimoku-v6
 229     =>>ruimoku-v6
 230     =jef-china3
 231     =shinjigen))
 232
 233
 234 ;;; @ char-db formatters
 235 ;;;
 236
 237 (defun char-db-make-char-spec (char)
 238   (let (ret char-spec)
 239     (cond ((characterp char)
 240            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 241                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 242                   (setq char-spec (list (cons '=ucs ret)))
 243                   (cond ((setq ret (get-char-attribute char 'name))
 244                          (setq char-spec (cons (cons 'name ret) char-spec))
 245                          )
 246                         ((setq ret (get-char-attribute char 'name*))
 247                          (setq char-spec (cons (cons 'name* ret) char-spec))
 248                          ))
 249                   )
 250                  ((setq ret
 251                         (catch 'tag
 252                           (let ((rest char-db-coded-charset-priority-list)
 253                                 ccs)
 254                             (while rest
 255                               (setq ccs (charset-name
 256                                          (find-charset (car rest))))
 257                               (if (setq ret
 258                                         (encode-char char ccs
 259                                                      'defined-only))
 260                                   (throw 'tag (cons ccs ret)))
 261                               (setq rest (cdr rest))))))
 262                   (setq char-spec (list ret))
 263                   (dolist (ccs (delq (car ret) (charset-list)))
 264                     (if (and (or (charset-iso-final-char ccs)
 265                                  (memq ccs
 266                                        '(=daikanwa
 267                                          =daikanwa@rev2
 268                                          ;; =gt-k
 269                                          =jis-x0208@1997
 270                                          ))
 271                                  (string-match "=ucs@" (symbol-name ccs)))
 272                              (setq ccs (charset-name ccs))
 273                              (null (assq ccs char-spec))
 274                              (setq ret (encode-char char ccs 'defined-only)))
 275                         (setq char-spec (cons (cons ccs ret) char-spec))))
 276                   (if (null char-spec)
 277                       (setq char-spec (split-char char)))
 278                   (cond ((setq ret (get-char-attribute char 'name))
 279                          (setq char-spec (cons (cons 'name ret) char-spec))
 280                          )
 281                         ((setq ret (get-char-attribute char 'name*))
 282                          (setq char-spec (cons (cons 'name* ret) char-spec))
 283                          ))
 284                   )
 285                  ((setq ret (get-char-attribute
 286                              char 'ideographic-combination))
 287                   (setq char-spec
 288                         (cons (cons 'ideographic-combination ret)
 289                               char-spec))
 290                   ))
 291            char-spec)
 292           ((consp char)
 293            char))))
 294
 295 (defun char-db-insert-char-spec (char &optional readable column
 296                                       required-features)
 297   (unless column
 298     (setq column (current-column)))
 299   (let (char-spec temp-char)
 300     (setq char-spec (char-db-make-char-spec char))
 301     (unless (or (characterp char) ; char
 302                 (condition-case nil
 303                     (setq char (find-char char-spec))
 304                   (error nil)))
 305       ;; define temporary character
 306       ;;   Current implementation is dirty.
 307       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 308                                          char-spec)))
 309       (remove-char-attribute temp-char 'ideograph-daikanwa)
 310       (setq char temp-char))
 311     (insert-char-attributes char
 312                             readable
 313                             (union (mapcar #'car char-spec)
 314                                    required-features)
 315                             nil 'for-sub-node)
 316     (when temp-char
 317       ;; undefine temporary character
 318       ;;   Current implementation is dirty.
 319       (setq char-spec (char-attribute-alist temp-char))
 320       (while char-spec
 321         (remove-char-attribute temp-char (car (car char-spec)))
 322         (setq char-spec (cdr char-spec))))))
 323
 324 (defun char-db-insert-alist (alist &optional readable column)
 325   (unless column
 326     (setq column (current-column)))
 327   (let ((line-breaking
 328          (concat "\n" (make-string (1+ column) ?\ )))
 329         name value
 330         ret al ; cal
 331         key
 332         lbs cell rest separator)
 333     (insert "(")
 334     (while alist
 335       (setq name (car (car alist))
 336             value (cdr (car alist)))
 337       (cond ((eq name 'char)
 338              (insert "(char . ")
 339              (if (setq ret (condition-case nil
 340                                (find-char value)
 341                              (error nil)))
 342                  (progn
 343                    (setq al nil
 344                          ;; cal nil
 345                          )
 346                    (while value
 347                      (setq key (car (car value)))
 348                      ;; (if (find-charset key)
 349                      ;;     (setq cal (cons key cal))
 350                      (setq al (cons key al))
 351                      ;; )
 352                      (setq value (cdr value)))
 353                    (insert-char-attributes ret
 354                                            readable
 355                                            (or al 'none) ; cal
 356                                            nil 'for-sub-node))
 357                (insert (prin1-to-string value)))
 358              (insert ")")
 359              (insert line-breaking))
 360             ((consp value)
 361              (insert (format "(%-18s " name))
 362              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 363              (while (consp value)
 364                (setq cell (car value))
 365                (if (and (consp cell)
 366                         (consp (car cell))
 367                         (setq ret (condition-case nil
 368                                       (find-char cell)
 369                                     (error nil)))
 370                         )
 371                    (progn
 372                      (setq rest cell
 373                            al nil
 374                            ;; cal nil
 375                            )
 376                      (while rest
 377                        (setq key (car (car rest)))
 378                        ;; (if (find-charset key)
 379                        ;;     (setq cal (cons key cal))
 380                        (setq al (cons key al))
 381                        ;; )
 382                        (setq rest (cdr rest)))
 383                      (if separator
 384                          (insert lbs))
 385                      (insert-char-attributes ret
 386                                              readable
 387                                              al ; cal
 388                                              nil 'for-sub-node)
 389                      (setq separator lbs))
 390                  (if separator
 391                      (insert separator))
 392                  (insert (prin1-to-string cell))
 393                  (setq separator " "))
 394                (setq value (cdr value)))
 395              (insert ")")
 396              (insert line-breaking))
 397             (t
 398              (insert (format "(%-18s . %S)%s"
 399                              name value
 400                              line-breaking))))
 401       (setq alist (cdr alist))))
 402   (insert ")"))
 403
 404 (defun char-db-insert-char-reference (plist &optional readable column)
 405   (unless column
 406     (setq column (current-column)))
 407   (let ((line-breaking
 408          (concat "\n" (make-string (1+ column) ?\ )))
 409         (separator "")
 410         name value)
 411     (insert "(")
 412     (while plist
 413       (setq name (pop plist))
 414       (setq value (pop plist))
 415       (cond ((eq name :char)
 416              (insert separator)
 417              (insert ":char\t")
 418              (cond ((numberp value)
 419                     (setq value (decode-char '=ucs value)))
 420                    ;; ((consp value)
 421                    ;;  (setq value (or (find-char value)
 422                    ;;                  value)))
 423                    )
 424              (char-db-insert-char-spec value readable)
 425              (insert line-breaking)
 426              (setq separator ""))
 427             ((eq name :radical)
 428              (insert (format "%s%s\t%d ; %c%s"
 429                              separator
 430                              name value
 431                              (ideographic-radical value)
 432                              line-breaking))
 433              (setq separator ""))
 434             (t
 435              (insert (format "%s%s\t%S" separator name value))
 436              (setq separator line-breaking)))
 437       ))
 438   (insert ")"))
 439
 440 (defun char-db-decode-isolated-char (ccs code-point)
 441   (let (ret)
 442     (setq ret
 443           (cond ((eq ccs 'arabic-iso8859-6)
 444                  (decode-char ccs code-point))
 445                 ((and (memq ccs '(=gt-pj-1
 446                                   =gt-pj-2
 447                                   =gt-pj-3
 448                                   =gt-pj-4
 449                                   =gt-pj-5
 450                                   =gt-pj-6
 451                                   =gt-pj-7
 452                                   =gt-pj-8
 453                                   =gt-pj-9
 454                                   =gt-pj-10
 455                                   =gt-pj-11))
 456                       (setq ret (decode-char ccs code-point))
 457                       (setq ret (encode-char ret '=gt 'defined-only)))
 458                  (decode-builtin-char '=gt ret))
 459                 (t
 460                  (decode-builtin-char ccs code-point))))
 461     (cond ((and (<= 0 (char-int ret))
 462                 (<= (char-int ret) #x1F))
 463            (decode-char '=ucs (+ #x2400 (char-int ret))))
 464           ((= (char-int ret) #x7F)
 465            ?\u2421)
 466           (t ret))))
 467
 468 (defvar char-db-convert-obsolete-format t)
 469
 470 (defun char-db-insert-ccs-feature (name value line-breaking)
 471   (cond
 472    ((integerp value)
 473     (insert
 474      (format
 475       (cond
 476        ((memq name '(=shinjigen
 477                      =shinjigen@1ed
 478                      =shinjigen@rev =shinjigen/+p@rev
 479                      =daikanwa/ho))
 480         "(%-18s .  %04d)\t; %c")
 481        ((eq name '=shinjigen@1ed/24pr)
 482         "(%-18s . %04d)\t; %c")
 483        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 484                          =daikanwa@rev1 =daikanwa@rev2
 485                          =daikanwa/+p =daikanwa/+2p
 486                          =gt =>>>gt =>>gt =>gt
 487                          =gt-k =>>gt-k =>gt-k
 488                          =>>adobe-japan1
 489                          =cbeta =>>cbeta
 490                          =zinbun-oracle =>zinbun-oracle))
 491             (string-match "^=adobe-" (symbol-name name)))
 492         "(%-18s . %05d)\t; %c")
 493        ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 494         "(%-18s . %06d)\t; %c")
 495        ((>= (charset-dimension name) 2)
 496         "(%-18s . #x%04X)\t; %c")
 497        (t
 498         "(%-18s . #x%02X)\t; %c"))
 499       name
 500       (if (= (charset-iso-graphic-plane name) 1)
 501           (logior value
 502                   (cond ((= (charset-dimension name) 1)
 503                          #x80)
 504                         ((= (charset-dimension name) 2)
 505                          #x8080)
 506                         ((= (charset-dimension name) 3)
 507                          #x808080)
 508                         (t 0)))
 509         value)
 510       (char-db-decode-isolated-char name value)))
 511     (if (and (= (charset-chars name) 94)
 512              (= (charset-dimension name) 2))
 513         (insert (format " [%02d-%02d]"
 514                         (- (lsh value -8) 32)
 515                         (- (logand value 255) 32))))
 516     )
 517    (t
 518     (insert (format "(%-18s . %s)" name value))
 519     ))
 520   (insert line-breaking))
 521
 522 (defun char-db-insert-relation-feature (char name value line-breaking
 523                                              ccss readable)
 524   (insert (format "(%-18s%s " name line-breaking))
 525   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 526         separator cell sources required-features
 527         ret)
 528     (while (consp value)
 529       (setq cell (car value))
 530       (if (integerp cell)
 531           (setq cell (decode-char '=ucs cell)))
 532       (cond
 533        ((eq name '->subsumptive)
 534         (when (or (not (some (lambda (atr)
 535                                (get-char-attribute cell atr))
 536                              char-db-ignored-attributes))
 537                   (some (lambda (ccs)
 538                           (encode-char cell ccs 'defined-only))
 539                         ccss))
 540           (if separator
 541               (insert lbs))
 542           (let ((char-db-ignored-attributes
 543                  (cons '<-subsumptive
 544                        char-db-ignored-attributes)))
 545             (insert-char-attributes cell readable nil nil 'for-sub-node))
 546           (setq separator lbs))
 547         )
 548        ((characterp cell)
 549         (setq sources
 550               (get-char-attribute
 551                char (intern (format "%s*sources" name))))
 552         (setq required-features nil)
 553         (dolist (source sources)
 554           (cond
 555            ((memq source '(JP
 556                            JP/Jouyou
 557                            shinjigen shinjigen@1ed shinjigen@rev))
 558             (setq required-features
 559                   (union required-features
 560                          '(=jis-x0208
 561                            =jis-x0208@1990
 562                            =jis-x0213-1@2000
 563                            =jis-x0213-1@2004
 564                            =jis-x0213-2
 565                            =jis-x0212
 566                            =jis-x0208@1983
 567                            =jis-x0208@1978
 568                            =shinjigen))))
 569            ((eq source 'CN)
 570             (setq required-features
 571                   (union required-features
 572                          '(=gb2312
 573                            =gb12345
 574                            =iso-ir165)))))
 575           (cond
 576            ((find-charset (setq ret (intern (format "=%s" source))))
 577             (setq required-features
 578                   (cons ret required-features)))
 579            (t (setq required-features
 580                     (cons source required-features)))))
 581         (cond ((string-match "@JP" (symbol-name name))
 582                (setq required-features
 583                      (union required-features
 584                             '(=jis-x0208
 585                               =jis-x0208@1990
 586                               =jis-x0213-1-2000
 587                               =jis-x0213-2-2000
 588                               =jis-x0212
 589                               =jis-x0208@1983
 590                               =jis-x0208@1978))))
 591               ((string-match "@CN" (symbol-name name))
 592                (setq required-features
 593                      (union required-features
 594                             '(=gb2312
 595                               =gb12345
 596                               =iso-ir165)))))
 597         (if separator
 598             (insert lbs))
 599         (if readable
 600             (insert (format "%S" cell))
 601           (char-db-insert-char-spec cell readable
 602                                     nil
 603                                     required-features))
 604         (setq separator lbs))
 605        ((consp cell)
 606         (if separator
 607             (insert lbs))
 608         (if (consp (car cell))
 609             (char-db-insert-char-spec cell readable)
 610           (char-db-insert-char-reference cell readable))
 611         (setq separator lbs))
 612        (t
 613         (if separator
 614             (insert separator))
 615         (insert (prin1-to-string cell))
 616         (setq separator " ")))
 617       (setq value (cdr value)))
 618     (insert ")")
 619     (insert line-breaking)))
 620
 621 (defun insert-char-attributes (char &optional readable attributes column
 622                                     for-sub-node)
 623   (unless column
 624     (setq column (current-column)))
 625   (let (name value ; has-long-ccs-name
 626         rest
 627         radical strokes
 628         (line-breaking
 629          (concat "\n" (make-string (1+ column) ?\ )))
 630         lbs cell separator ret
 631         key al cal
 632         dest-ccss ; sources required-features
 633         ccss)
 634     (let (atr-d)
 635       (setq attributes
 636             (sort (if attributes
 637                       (if (consp attributes)
 638                           (progn
 639                             (dolist (name attributes)
 640                               (unless (memq name char-db-ignored-attributes)
 641                                 (if (find-charset name)
 642                                     (push name ccss))
 643                                 (push name atr-d)))
 644                             atr-d))
 645                     (dolist (name (char-attribute-list))
 646                       (unless (memq name char-db-ignored-attributes)
 647                         (if (find-charset name)
 648                             (push name ccss))
 649                         (push name atr-d)))
 650                     atr-d)
 651                   #'char-attribute-name<)))
 652     (insert "(")
 653     (when (memq '<-subsumptive attributes)
 654       (when (or readable (not for-sub-node))
 655         (when (setq value (get-char-attribute char '<-subsumptive))
 656           (char-db-insert-relation-feature char '<-subsumptive value
 657                                            line-breaking
 658                                            ccss readable)))
 659       (setq attributes (delq '<-subsumptive attributes)))
 660     (when (and (memq '<-denotational attributes)
 661                (setq value (get-char-attribute char '<-denotational)))
 662       (char-db-insert-relation-feature char '<-denotational value
 663                                        line-breaking
 664                                        ccss readable)
 665       (setq attributes (delq '<-denotational attributes)))
 666     (when (and (memq 'name attributes)
 667                (setq value (get-char-attribute char 'name)))
 668       (insert (format
 669                (if (> (+ (current-column) (length value)) 48)
 670                    "(name . %S)%s"
 671                  "(name               . %S)%s")
 672                value line-breaking))
 673       (setq attributes (delq 'name attributes))
 674       )
 675     (when (and (memq 'name* attributes)
 676                (setq value (get-char-attribute char 'name*)))
 677       (insert (format
 678                (if (> (+ (current-column) (length value)) 48)
 679                    "(name* . %S)%s"
 680                  "(name*              . %S)%s")
 681                value line-breaking))
 682       (setq attributes (delq 'name* attributes))
 683       )
 684     (when (and (memq 'script attributes)
 685                (setq value (get-char-attribute char 'script)))
 686       (insert (format "(script\t\t%s)%s"
 687                       (mapconcat (function prin1-to-string)
 688                                  value " ")
 689                       line-breaking))
 690       (setq attributes (delq 'script attributes))
 691       )
 692     (dolist (name '(=>ucs =>ucs*))
 693       (when (and (memq name attributes)
 694                  (setq value (get-char-attribute char name)))
 695         (insert (format "(%-18s . #x%04X)\t; %c%s"
 696                         name value (decode-char '=ucs value)
 697                         line-breaking))
 698         (setq attributes (delq name attributes))))
 699     (dolist (name '(=>ucs@gb =>ucs@big5))
 700       (when (and (memq name attributes)
 701                  (setq value (get-char-attribute char name)))
 702         (insert (format "(%-18s . #x%04X)\t; %c%s"
 703                         name value
 704                         (decode-char (intern
 705                                       (concat "="
 706                                               (substring
 707                                                (symbol-name name) 2)))
 708                                      value)
 709                         line-breaking))
 710         (setq attributes (delq name attributes))
 711         ))
 712     ;; (dolist (name '(=>daikanwa))
 713     ;;   (when (and (memq name attributes)
 714     ;;              (setq value (get-char-attribute char name)))
 715     ;;     (insert
 716     ;;      (if (integerp value)
 717     ;;          (format "(%-18s . %05d)\t; %c%s"
 718     ;;                  name value (decode-char '=daikanwa value)
 719     ;;                  line-breaking)
 720     ;;        (format "(%-18s %s)\t; %c%s"
 721     ;;                name
 722     ;;                (mapconcat (function prin1-to-string)
 723     ;;                           value " ")
 724     ;;                (char-representative-of-daikanwa char)
 725     ;;                line-breaking)))
 726     ;;     (setq attributes (delq name attributes))))
 727     (when (and (memq 'general-category attributes)
 728                (setq value (get-char-attribute char 'general-category)))
 729       (insert (format
 730                "(general-category\t%s) ; %s%s"
 731                (mapconcat (lambda (cell)
 732                             (format "%S" cell))
 733                           value " ")
 734                (cond ((rassoc value unidata-normative-category-alist)
 735                       "Normative Category")
 736                      ((rassoc value unidata-informative-category-alist)
 737                       "Informative Category")
 738                      (t
 739                       "Unknown Category"))
 740                line-breaking))
 741       (setq attributes (delq 'general-category attributes))
 742       )
 743     (when (and (memq 'bidi-category attributes)
 744                (setq value (get-char-attribute char 'bidi-category)))
 745       (insert (format "(bidi-category\t. %S)%s"
 746                       value
 747                       line-breaking))
 748       (setq attributes (delq 'bidi-category attributes))
 749       )
 750     (unless (or (not (memq 'mirrored attributes))
 751                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 752                     'empty))
 753       (insert (format "(mirrored\t\t. %S)%s"
 754                       value
 755                       line-breaking))
 756       (setq attributes (delq 'mirrored attributes))
 757       )
 758     (cond
 759      ((and (memq 'decimal-digit-value attributes)
 760            (setq value (get-char-attribute char 'decimal-digit-value)))
 761       (insert (format "(decimal-digit-value . %S)%s"
 762                       value
 763                       line-breaking))
 764       (setq attributes (delq 'decimal-digit-value attributes))
 765       (when (and (memq 'digit-value attributes)
 766                  (setq value (get-char-attribute char 'digit-value)))
 767         (insert (format "(digit-value\t . %S)%s"
 768                         value
 769                         line-breaking))
 770         (setq attributes (delq 'digit-value attributes))
 771         )
 772       (when (and (memq 'numeric-value attributes)
 773                  (setq value (get-char-attribute char 'numeric-value)))
 774         (insert (format "(numeric-value\t . %S)%s"
 775                         value
 776                         line-breaking))
 777         (setq attributes (delq 'numeric-value attributes))
 778         )
 779       )
 780      (t
 781       (when (and (memq 'digit-value attributes)
 782                  (setq value (get-char-attribute char 'digit-value)))
 783         (insert (format "(digit-value\t. %S)%s"
 784                         value
 785                         line-breaking))
 786         (setq attributes (delq 'digit-value attributes))
 787         )
 788       (when (and (memq 'numeric-value attributes)
 789                  (setq value (get-char-attribute char 'numeric-value)))
 790         (insert (format "(numeric-value\t. %S)%s"
 791                         value
 792                         line-breaking))
 793         (setq attributes (delq 'numeric-value attributes))
 794         )))
 795     (when (and (memq 'iso-10646-comment attributes)
 796                (setq value (get-char-attribute char 'iso-10646-comment)))
 797       (insert (format "(iso-10646-comment\t. %S)%s"
 798                       value
 799                       line-breaking))
 800       (setq attributes (delq 'iso-10646-comment attributes))
 801       )
 802     (when (and (memq 'morohashi-daikanwa attributes)
 803                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 804       (insert (format "(morohashi-daikanwa\t%s)%s"
 805                       (mapconcat (function prin1-to-string) value " ")
 806                       line-breaking))
 807       (setq attributes (delq 'morohashi-daikanwa attributes))
 808       )
 809     (setq radical nil
 810           strokes nil)
 811     (when (and (memq 'ideographic-radical attributes)
 812                (setq value (get-char-attribute char 'ideographic-radical)))
 813       (setq radical value)
 814       (insert (format "(ideographic-radical . %S)\t; %c%s"
 815                       radical
 816                       (ideographic-radical radical)
 817                       line-breaking))
 818       (setq attributes (delq 'ideographic-radical attributes))
 819       )
 820     (when (and (memq 'shuowen-radical attributes)
 821                (setq value (get-char-attribute char 'shuowen-radical)))
 822       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 823                       value
 824                       (shuowen-radical value)
 825                       line-breaking))
 826       (setq attributes (delq 'shuowen-radical attributes))
 827       )
 828     (let (key)
 829       (dolist (domain
 830                (append
 831                 char-db-feature-domains
 832                 (let (dest domain)
 833                   (dolist (feature (char-attribute-list))
 834                     (setq feature (symbol-name feature))
 835                     (when (string-match
 836                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 837                            feature)
 838                       (setq domain (intern (match-string 2 feature)))
 839                      (unless (memq domain dest)
 840                        (setq dest (cons domain dest)))))
 841                   (sort dest #'string<))))
 842         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 843         (when (and (memq key attributes)
 844                    (setq value (get-char-attribute char key)))
 845           (setq radical value)
 846           (insert (format "(%s . %S)\t; %c%s"
 847                           key
 848                           radical
 849                           (ideographic-radical radical)
 850                           line-breaking))
 851           (setq attributes (delq key attributes))
 852           )
 853         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 854         (when (and (memq key attributes)
 855                    (setq value (get-char-attribute char key)))
 856           (setq strokes value)
 857           (insert (format "(%s . %S)%s"
 858                           key
 859                           strokes
 860                           line-breaking))
 861           (setq attributes (delq key attributes))
 862           )
 863         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 864         (when (and (memq key attributes)
 865                    (setq value (get-char-attribute char key)))
 866           (insert (format "(%s       . %S)%s"
 867                           key
 868                           value
 869                           line-breaking))
 870           (setq attributes (delq key attributes))
 871           )
 872         (dolist (feature '(ideographic-radical
 873                            ideographic-strokes
 874                            total-strokes))
 875           (setq key (intern (format "%s@%s*sources" feature domain)))
 876           (when (and (memq key attributes)
 877                      (setq value (get-char-attribute char key)))
 878             (insert (format "(%s%s" key line-breaking))
 879             (dolist (cell value)
 880               (insert (format " %s" cell)))
 881             (insert ")")
 882             (insert line-breaking)
 883             (setq attributes (delq key attributes))
 884             ))
 885         ))
 886     (when (and (memq 'ideographic-strokes attributes)
 887                (setq value (get-char-attribute char 'ideographic-strokes)))
 888       (setq strokes value)
 889       (insert (format "(ideographic-strokes . %S)%s"
 890                       strokes
 891                       line-breaking))
 892       (setq attributes (delq 'ideographic-strokes attributes))
 893       )
 894     (when (and (memq 'kangxi-radical attributes)
 895                (setq value (get-char-attribute char 'kangxi-radical)))
 896       (unless (eq value radical)
 897         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 898                         value
 899                         (ideographic-radical value)
 900                         line-breaking))
 901         (or radical
 902             (setq radical value)))
 903       (setq attributes (delq 'kangxi-radical attributes))
 904       )
 905     (when (and (memq 'kangxi-strokes attributes)
 906                (setq value (get-char-attribute char 'kangxi-strokes)))
 907       (unless (eq value strokes)
 908         (insert (format "(kangxi-strokes\t . %S)%s"
 909                         value
 910                         line-breaking))
 911         (or strokes
 912             (setq strokes value)))
 913       (setq attributes (delq 'kangxi-strokes attributes))
 914       )
 915     (when (and (memq 'japanese-radical attributes)
 916                (setq value (get-char-attribute char 'japanese-radical)))
 917       (unless (eq value radical)
 918         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 919                         value
 920                         (ideographic-radical value)
 921                         line-breaking))
 922         (or radical
 923             (setq radical value)))
 924       (setq attributes (delq 'japanese-radical attributes))
 925       )
 926     (when (and (memq 'japanese-strokes attributes)
 927                (setq value (get-char-attribute char 'japanese-strokes)))
 928       (unless (eq value strokes)
 929         (insert (format "(japanese-strokes\t . %S)%s"
 930                         value
 931                         line-breaking))
 932         (or strokes
 933             (setq strokes value)))
 934       (setq attributes (delq 'japanese-strokes attributes))
 935       )
 936     (when (and (memq 'cns-radical attributes)
 937                (setq value (get-char-attribute char 'cns-radical)))
 938       (insert (format "(cns-radical\t . %S)\t; %c%s"
 939                       value
 940                       (ideographic-radical value)
 941                       line-breaking))
 942       (setq attributes (delq 'cns-radical attributes))
 943       )
 944     (when (and (memq 'cns-strokes attributes)
 945                (setq value (get-char-attribute char 'cns-strokes)))
 946       (unless (eq value strokes)
 947         (insert (format "(cns-strokes\t . %S)%s"
 948                         value
 949                         line-breaking))
 950         (or strokes
 951             (setq strokes value)))
 952       (setq attributes (delq 'cns-strokes attributes))
 953       )
 954     (when (and (memq 'shinjigen-1-radical attributes)
 955                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 956       (unless (eq value radical)
 957         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 958                         value
 959                         (ideographic-radical value)
 960                         line-breaking))
 961         (or radical
 962             (setq radical value)))
 963       (setq attributes (delq 'shinjigen-1-radical attributes))
 964       )
 965     (when (and (memq 'ideographic- attributes)
 966                (setq value (get-char-attribute char 'ideographic-)))
 967       (insert "(ideographic-       ")
 968       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 969             separator nil)
 970       (while (consp value)
 971         (setq cell (car value))
 972         (if (integerp cell)
 973             (setq cell (decode-char '=ucs cell)))
 974         (cond ((characterp cell)
 975                (if separator
 976                    (insert lbs))
 977                (if readable
 978                    (insert (format "%S" cell))
 979                  (char-db-insert-char-spec cell readable))
 980                (setq separator lbs))
 981               ((consp cell)
 982                (if separator
 983                    (insert lbs))
 984                (if (consp (car cell))
 985                    (char-db-insert-char-spec cell readable)
 986                  (char-db-insert-char-reference cell readable))
 987                (setq separator lbs))
 988               (t
 989                (if separator
 990                    (insert separator))
 991                (insert (prin1-to-string cell))
 992                (setq separator " ")))
 993         (setq value (cdr value)))
 994       (insert ")")
 995       (insert line-breaking)
 996       (setq attributes (delq 'ideographic- attributes)))
 997     (when (and (memq 'total-strokes attributes)
 998                (setq value (get-char-attribute char 'total-strokes)))
 999       (insert (format "(total-strokes       . %S)%s"
1000                       value
1001                       line-breaking))
1002       (setq attributes (delq 'total-strokes attributes))
1003       )
1004     (when (and (memq '->ideograph attributes)
1005                (setq value (get-char-attribute char '->ideograph)))
1006       (insert (format "(->ideograph\t%s)%s"
1007                       (mapconcat (lambda (code)
1008                                    (cond ((symbolp code)
1009                                           (symbol-name code))
1010                                          ((integerp code)
1011                                           (format "#x%04X" code))
1012                                          (t
1013                                           (format "%s %S"
1014                                                   line-breaking code))))
1015                                  value " ")
1016                       line-breaking))
1017       (setq attributes (delq '->ideograph attributes))
1018       )
1019     ;; (when (and (memq '->decomposition attributes)
1020     ;;            (setq value (get-char-attribute char '->decomposition)))
1021     ;;   (insert (format "(->decomposition\t%s)%s"
1022     ;;                   (mapconcat (lambda (code)
1023     ;;                                (cond ((symbolp code)
1024     ;;                                       (symbol-name code))
1025     ;;                                      ((characterp code)
1026     ;;                                       (if readable
1027     ;;                                           (format "%S" code)
1028     ;;                                         (format "#x%04X"
1029     ;;                                                 (char-int code))
1030     ;;                                         ))
1031     ;;                                      ((integerp code)
1032     ;;                                       (format "#x%04X" code))
1033     ;;                                      (t
1034     ;;                                       (format "%s%S" line-breaking code))))
1035     ;;                              value " ")
1036     ;;                   line-breaking))
1037     ;;   (setq attributes (delq '->decomposition attributes))
1038     ;;   )
1039     (if (equal (get-char-attribute char '->titlecase)
1040                (get-char-attribute char '->uppercase))
1041         (setq attributes (delq '->titlecase attributes)))
1042     (when (and (memq '->mojikyo attributes)
1043                (setq value (get-char-attribute char '->mojikyo)))
1044       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1045                       value (decode-char 'mojikyo value)
1046                       line-breaking))
1047       (setq attributes (delq '->mojikyo attributes))
1048       )
1049     (when (and (memq 'hanyu-dazidian-vol attributes)
1050                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1051       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1052                       value line-breaking))
1053       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1054       )
1055     (when (and (memq 'hanyu-dazidian-page attributes)
1056                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1057       (insert (format "(hanyu-dazidian-page . %d)%s"
1058                       value line-breaking))
1059       (setq attributes (delq 'hanyu-dazidian-page attributes))
1060       )
1061     (when (and (memq 'hanyu-dazidian-char attributes)
1062                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1063       (insert (format "(hanyu-dazidian-char . %d)%s"
1064                       value line-breaking))
1065       (setq attributes (delq 'hanyu-dazidian-char attributes))
1066       )
1067     (unless readable
1068       (dolist (ignored '(composition
1069                          ->denotational <-subsumptive ->ucs-unified
1070                          ->ideographic-component-forms))
1071         (setq attributes (delq ignored attributes))))
1072     (while attributes
1073       (setq name (car attributes))
1074       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1075                   'value-is-empty)
1076         (cond ((setq ret (find-charset name))
1077                (setq name (charset-name ret))
1078                (when (not (memq name dest-ccss))
1079                  (setq dest-ccss (cons name dest-ccss))
1080                  (char-db-insert-ccs-feature name value line-breaking))
1081                )
1082               ((string-match "^=>ucs@" (symbol-name name))
1083                (insert (format "(%-18s . #x%04X)\t; %c%s"
1084                                name value (decode-char '=ucs value)
1085                                line-breaking))
1086                )
1087               ((eq name 'jisx0208-1978/4X)
1088                (insert (format "(%-18s . #x%04X)%s"
1089                                name value
1090                                line-breaking))
1091                )
1092               ((and
1093                 (not readable)
1094                 (not (eq name '->subsumptive))
1095                 (not (eq name '->uppercase))
1096                 (not (eq name '->lowercase))
1097                 (not (eq name '->titlecase))
1098                 (not (eq name '->canonical))
1099                 (not (eq name '->Bopomofo))
1100                 (not (eq name '->mistakable))
1101                 (not (eq name '->ideographic-variants))
1102                 (null (get-char-attribute
1103                        char (intern (format "%s*sources" name))))
1104                 (not (string-match "\\*sources$" (symbol-name name)))
1105                 (null (get-char-attribute
1106                        char (intern (format "%s*note" name))))
1107                 (not (string-match "\\*note$" (symbol-name name)))
1108                 (or (eq name '<-identical)
1109                     (eq name '<-uppercase)
1110                     (eq name '<-lowercase)
1111                     (eq name '<-titlecase)
1112                     (eq name '<-canonical)
1113                     (eq name '<-ideographic-variants)
1114                     ;; (eq name '<-synonyms)
1115                     (string-match "^<-synonyms" (symbol-name name))
1116                     (eq name '<-mistakable)
1117                     (when (string-match "^->" (symbol-name name))
1118                       (cond
1119                        ((string-match "^->fullwidth" (symbol-name name))
1120                         (not (and (consp value)
1121                                   (characterp (car value))
1122                                   (encode-char
1123                                    (car value) '=ucs 'defined-only)))
1124                         )
1125                        (t)))
1126                     ))
1127                )
1128               ((or (eq name 'ideographic-structure)
1129                    (eq name 'ideographic-combination)
1130                    (eq name 'ideographic-)
1131                    (eq name '=decomposition)
1132                    (char-feature-base-name= '=decomposition name)
1133                    (char-feature-base-name= '=>decomposition name)
1134                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1135                    ;;               (symbol-name name))
1136                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1137                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1138                                  (symbol-name name))
1139                    )
1140                (char-db-insert-relation-feature char name value
1141                                                 line-breaking
1142                                                 ccss readable))
1143               ((memq name '(ideograph=
1144                             original-ideograph-of
1145                             ancient-ideograph-of
1146                             vulgar-ideograph-of
1147                             wrong-ideograph-of
1148                             ;; simplified-ideograph-of
1149                             ideographic-variants
1150                             ;; ideographic-different-form-of
1151                             ))
1152                (insert (format "(%-18s%s " name line-breaking))
1153                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1154                      separator nil)
1155                (while (consp value)
1156                  (setq cell (car value))
1157                  (if (and (consp cell)
1158                           (consp (car cell)))
1159                      (progn
1160                        (if separator
1161                            (insert lbs))
1162                        (char-db-insert-alist cell readable)
1163                        (setq separator lbs))
1164                    (if separator
1165                        (insert separator))
1166                    (insert (prin1-to-string cell))
1167                    (setq separator " "))
1168                  (setq value (cdr value)))
1169                (insert ")")
1170                (insert line-breaking))
1171               ((consp value)
1172                (insert (format "(%-18s " name))
1173                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1174                      separator nil)
1175                (while (consp value)
1176                  (setq cell (car value))
1177                  (if (and (consp cell)
1178                           (consp (car cell))
1179                           (setq ret (condition-case nil
1180                                         (find-char cell)
1181                                       (error nil))))
1182                      (progn
1183                        (setq rest cell
1184                              al nil
1185                              cal nil)
1186                        (while rest
1187                          (setq key (car (car rest)))
1188                          (if (find-charset key)
1189                              (setq cal (cons key cal))
1190                            (setq al (cons key al)))
1191                          (setq rest (cdr rest)))
1192                        (if separator
1193                            (insert lbs))
1194                        (insert-char-attributes ret
1195                                                readable
1196                                                al ; cal
1197                                                nil 'for-sub-node)
1198                        (setq separator lbs))
1199                    (setq ret (prin1-to-string cell))
1200                    (if separator
1201                        (if (< (+ (current-column)
1202                                  (length ret)
1203                                  (length separator))
1204                               76)
1205                            (insert separator)
1206                          (insert lbs)))
1207                    (insert ret)
1208                    (setq separator " "))
1209                  (setq value (cdr value)))
1210                (insert ")")
1211                (insert line-breaking))
1212               (t
1213                (insert (format "(%-18s" name))
1214                (setq ret (prin1-to-string value))
1215                (unless (< (+ (current-column)
1216                              (length ret)
1217                              3)
1218                           76)
1219                  (insert line-breaking))
1220                (insert " . " ret ")" line-breaking)
1221                ;; (insert (format "(%-18s . %S)%s"
1222                ;;                 name value
1223                ;;                 line-breaking))
1224                )
1225               ))
1226       (setq attributes (cdr attributes)))
1227     (insert ")")))
1228
1229 (defun insert-char-data (char &optional readable
1230                               attributes)
1231   (save-restriction
1232     (narrow-to-region (point)(point))
1233     (insert "(define-char
1234   '")
1235     (insert-char-attributes char readable attributes)
1236     (insert ")\n")
1237     (goto-char (point-min))
1238     (while (re-search-forward "[ \t]+$" nil t)
1239       (replace-match ""))
1240     ;; from tabify.
1241     (goto-char (point-min))
1242     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1243       (let ((column (current-column))
1244             (indent-tabs-mode t))
1245         (delete-region (match-beginning 0) (point))
1246         (indent-to column)))
1247     (goto-char (point-max))
1248     ;; (tabify (point-min)(point-max))
1249     ))
1250
1251 (defun insert-char-data-with-variant (char &optional printable
1252                                            no-ucs-unified
1253                                            script excluded-script)
1254   (insert-char-data char printable)
1255   (let ((variants (char-variants char))
1256         rest
1257         variant vs ret)
1258     (setq variants (sort variants #'<))
1259     (setq rest variants)
1260     (setq variants (cons char variants))
1261     (while rest
1262       (setq variant (car rest))
1263       (unless (get-char-attribute variant '<-subsumptive)
1264         (if (and (or (null script)
1265                      (null (setq vs (get-char-attribute variant 'script)))
1266                      (memq script vs))
1267                  (or (null excluded-script)
1268                      (null (setq vs (get-char-attribute variant 'script)))
1269                      (not (memq excluded-script vs))))
1270             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1271               (insert-char-data variant printable)
1272               (if (setq ret (char-variants variant))
1273                   (while ret
1274                     (or (memq (car ret) variants)
1275                         ;; (get-char-attribute (car ret) '<-subsumptive)
1276                         (setq rest (nconc rest (list (car ret)))))
1277                     (setq ret (cdr ret)))))))
1278       (setq rest (cdr rest)))))
1279
1280 (defun insert-char-range-data (min max &optional script excluded-script)
1281   (let ((code min)
1282         char)
1283     (while (<= code max)
1284       (setq char (decode-char '=ucs code))
1285       (if (encode-char char '=ucs 'defined-only)
1286           (insert-char-data-with-variant char nil 'no-ucs-unified
1287                                          script excluded-script))
1288       (setq code (1+ code)))))
1289
1290 (defun write-char-range-data-to-file (min max file
1291                                           &optional script excluded-script)
1292   (let ((coding-system-for-write char-db-file-coding-system))
1293     (with-temp-buffer
1294       (insert (format ";; -*- coding: %s -*-\n"
1295                       char-db-file-coding-system))
1296       (insert-char-range-data min max script excluded-script)
1297       (write-region (point-min)(point-max) file))))
1298
1299 (defvar what-character-original-window-configuration)
1300
1301 ;;;###autoload
1302 (defun what-char-definition (char)
1303   (interactive (list (char-after)))
1304   (let ((buf (get-buffer-create "*Character Description*"))
1305         (the-buf (current-buffer))
1306         (win-conf (current-window-configuration)))
1307     (pop-to-buffer buf)
1308     (make-local-variable 'what-character-original-window-configuration)
1309     (setq what-character-original-window-configuration win-conf)
1310     (setq buffer-read-only nil)
1311     (erase-buffer)
1312     (condition-case err
1313         (progn
1314           (insert-char-data-with-variant char 'printable)
1315           (unless (char-attribute-alist char)
1316             (insert (format ";; = %c\n"
1317                             (let* ((rest (split-char char))
1318                                    (ccs (pop rest))
1319                                    (code (pop rest)))
1320                               (while rest
1321                                 (setq code (logior (lsh code 8)
1322                                                    (pop rest))))
1323                               (decode-char ccs code)))))
1324           ;; (char-db-update-comment)
1325           (set-buffer-modified-p nil)
1326           (view-mode the-buf (lambda (buf)
1327                                (set-window-configuration
1328                                 what-character-original-window-configuration)
1329                                ))
1330           (goto-char (point-min)))
1331       (error (progn
1332                (set-window-configuration
1333                 what-character-original-window-configuration)
1334                (signal (car err) (cdr err)))))))
1335
1336
1337 ;;; @ end
1338 ;;;
1339
1340 (provide 'char-db-util)
1341
1342 ;;; char-db-util.el ends here